BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:CBL Alumni Talk: Latent Stochastic Differential Equations: An Unex
 plored Model Class. - David Duvenaud\, University of Toronto
DTSTART:20210604T150000Z
DTEND:20210604T160000Z
UID:TALK158827@talks.cam.ac.uk
CONTACT:Elre Oldewage
DESCRIPTION:We show how to do gradient-based stochastic variational infere
 nce in stochastic differential equations (SDEs)\, in a way that allows the
  use of adaptive SDE solvers.  This allows us to scalably fit a new family
  of richly-parameterized distributions over irregularly-sampled time serie
 s.  We apply latent SDEs to motion capture data\, and to demonstrate infin
 itely-deep Bayesian neural networks.  We also discuss the pros and cons of
  this barely-explored model class\, comparing it to Gaussian processes and
  neural processes.\n\nSome technical details are in this paper: https://ar
 xiv.org/abs/2001.01328\nAnd code is available at: https://github.com/googl
 e-research/torchsde\n\nBio: David Duvenaud is an assistant professor in co
 mputer science at the University of Toronto. His research focuses on conti
 nuous-time models\, latent-variable models\, and deep learning.  His postd
 oc was done at Harvard University\, and his Ph.D. at the University of Cam
 bridge.  David also co-founded Invenia\, an energy forecasting company.
LOCATION:https://eng-cam.zoom.us/j/84495932262?pwd=MlFJL3Z3c1JmenFOY2xJQTN
 PSzdsQT09
END:VEVENT
END:VCALENDAR